Verse 22 stated the bare datum from Genesis: Abraham had two sons. Verse 23 supplies the asymmetry that makes the datum useful. The two sons are not two comparable cases; they stand under two different orders of generation. One birth is κατὰ σάρκα, in line with ordinary human procreation. The other is διʼ ἐπαγγελίας, brought about through a promise.

Three features carry the weight:

1. The verb occurs once. γεγέννηται governs both clauses; the second is verbless and borrows it. So Paul is not contrasting two different verbs of birth but one birth-verb qualified two ways.
2. The prepositions are deliberately unmatched. κατά + accusative gives a norm or standard; διά + genitive gives means. Paul did not write κατὰ ἐπαγγελίαν. The flesh is the pattern the first birth conformed to; the promise is the instrument that produced the second.
3. The perfect tense. γεγέννηται is not the aorist ἐγεννήθη. Two births roughly two millennia old are presented as facts whose results still stand — which is precisely what allows the Galatian present tense of v. 29 (‘so also now’).

The mothers are identified only by legal status: ἡ παιδίσκη and ἡ ἐλευθέρα. Neither Hagar nor Sarah is named at this point (Hagar appears at v. 24, Sarah never). Paul is arguing from slavery and freedom, not from personalities.

ἀλλʼ here is properly adversative, not ascensive. Verse 22 could be read as establishing an equivalence — two sons of one father — and ἀλλά cuts that off. Per the corpus policy, where no adjunctive καί or μενοῦνγε follows and the clause qualifies rather than heightens, ἀλλά is rendered plainly ‘but’ and kept clause-initial.

The μέν … δέ correlation then divides the pair. Per §10, epistolary correlatives are argument structure and are rendered, not dropped. English cannot carry μέν with a single word without stiffness, so the pairing is marked here by ‘But the one … while the one …’, which reproduces the balance without inventing an adversative that δέ alone does not assert.

Note also the shift in construction from v. 22: there Paul wrote ἕνα ἐκ τῆς παιδίσκης (‘one out of the slave-girl’), a bare numeral; here he writes ὁ ἐκ τῆς παιδίσκης with the article. The article converts the prepositional phrase into a defining modifier — this is the son characterized by that origin, not merely one of two.

Built onbdf · na28

γεννάω in the passive is the standard verb for being begotten or born. The form here is perfect passive indicative, and both the tense and the voice matter.

Tense. The perfect denotes a completed act with a result that stands. Against the aorist (ἐγεννήθη, which Paul does use at v. 29 of the same person), the perfect keeps the consequence in view: Ishmael’s fleshly derivation is not merely a historical event but a permanent classification. Some commentators treat the perfect as loosely aoristic in Koine, and that possibility is real — but the presence of the aorist γεννηθείς at v. 29 within the same argument suggests Paul was capable of the distinction and made it here on purpose.

Ellipsis. The verb is not repeated in the second clause. Greek freely elides an identical predicate across a μέν … δέ pair, and English must supply something. The literal tier brackets the supplied verb; the reading text repeats ‘has been born’ so that the reader can see one verb qualified twice rather than two different assertions.

Divine passive. Whether the second clause’s implied passive has God as unexpressed agent is a live question. The instrumental διʼ ἐπαγγελίας names the means and leaves the agent unnamed, which is characteristic of Paul’s reticence; but nothing in the syntax forces a theological passive, and Sarah’s own bearing of Isaac is a plain historical fact. The English is left agentless.

Built onbdag · bdf

παιδίσκη is a diminutive of παῖς and in the LXX is the regular equivalent of Hebrew שִׁפְחָה / אָמָה — the female slave. It is the word Genesis 16:1-3 LXX uses of Hagar, and Gen 21:10 LXX uses it in the very sentence Paul will quote at v. 30.

Per the standing corpus decision on δοῦλος, the ownership dimension is retained rather than softened: this is a slave woman, not a ‘maid’ or ‘servant girl’ in the domestic-employee sense modern English hears. The whole allegory turns on legal non-freedom (δουλεία, v. 24; δουλεύει, v. 25), and blunting the noun here quietly disarms vv. 24-31.

‘Slave-girl’ is retained in the literal tier because the diminutive is morphologically real; ‘slave woman’ is used in the reading text, since English ‘girl’ now carries an age implication the Greek diminutive has largely lost by the Koine period.

Built onbdag · muraoka-lxx

ἡ ἐλευθέρα is a substantival adjective: ‘the free [woman]’. Like παιδίσκη it is a status term drawn from Gen 21:10 LXX, and its force is juridical — the free woman’s son inherits, the slave’s does not. That legal principle, not any moral comparison of the two women, is what Paul is exploiting.

Sarah is not named anywhere in Galatians 4. This is worth registering: Paul names Hagar (v. 24) because he needs her for the Sinai identification, but he leaves the free woman unnamed so that she can absorb the identification with the Jerusalem above (v. 26). Supplying ‘Sarah’ in the rendering would close that opening.

Built onbdag · muraoka-lxx

The two qualifiers are not grammatically parallel, and the asymmetry is the argument.

κατὰ σάρκα (accusative) is the norm-or-standard use of κατά, consistent with the corpus policy on κατά + accusative: ‘according to, measured by’. Ishmael’s birth conformed to the ordinary pattern of human generation. The phrase is Paul’s fixed idiom (Rom 1:3; 4:1; 9:3, 5; 1 Cor 1:26; 10:18; 2 Cor 5:16; Gal 4:29), and in most of those places it is descriptive rather than condemnatory.

διʼ ἐπαγγελίας (genitive) is instrumental: the promise is the means through which Isaac came to be. Paul had κατὰ ἐπαγγελίαν available — he uses exactly that phrase at Gal 3:29 — and chose otherwise here. The switch is data.

Built onbdag · bdf

ἐπαγγελία in classical usage is an announcement or a public undertaking; in the LXX and NT the theologically loaded sense of divine promise dominates, and in Galatians it is a keyword — 3:14, 16, 17, 18 (twice), 21, 22, 29; 4:23, 28.

It is anarthrous here, hence qualitative: ‘through promise’ as a mode of operation, over against ‘through flesh’ as a mode. English reads awkwardly without an article, so the reading text supplies an indefinite one and the literal tier leaves it bare. The specific referent is Gen 17:16 / 18:10 / 21:1-2 — the word spoken to Abraham and Sarah, on which Isaac’s very existence depended.

The diachronic point matters (§9): ἐπαγγελία is not a prediction that happened to come true. In Galatians it is a legally weighted undertaking — Paul treats it in 3:15-18 on the analogy of a ratified διαθήκη. The birth ‘through promise’ is therefore a birth grounded in a binding divine commitment, which is why it can support an inheritance claim in v. 30.

Built onbdag · muraoka-lxx

ὁ ἐκ τῆς παιδίσκης — article plus prepositional phrase, substantivized. The construction is the same one the corpus has already registered at Phil 4:22 (οἱ ἐκ τῆς Καίσαρος οἰκίας) and Gal 3:7 (οἱ ἐκ πίστεως): the article converts the phrase into a definite substantive, and ἐκ marks derivation rather than motion away.

The elided noun is υἱός, carried over from v. 22. The literal tier shows ‘the one out of the slave-girl’ so the construction stays visible; the reading text supplies ‘the son’, since English otherwise reads ‘the one’ as a vague demonstrative rather than as a defined member of the pair.

Note the Greek keeps the masculine article against the feminine genitives — the sons are being classified by their mothers, and the grammar keeps both parties in view in a way that a smoothed English ‘Ishmael … Isaac’ would erase. Paul does not name either son in this verse either.

Built onbdf · bdag

Verse 23 does not quote Genesis; it summarizes it in Paul’s own vocabulary while keeping the LXX’s two status-nouns.

• παιδίσκη is Hagar’s designation at Gen 16:1-3; 21:10, 12, 13 LXX.
• ἐλευθέρα is not a Genesis word. Paul supplies it as the logical complement of παιδίσκη, drawing on the legal antithesis rather than the narrative’s own phrasing.
• The promise-birth is Gen 17:15-21 and 18:10-14, sealed at 21:1-2: ‘YHWH did for Sarah as he had spoken.’ Sarah’s barrenness and age make the point Paul needs — Isaac’s existence traces to a spoken word, not to reproductive capacity.

One consequence of Paul’s compression: Genesis itself calls Ishmael Abraham’s son and records God’s blessing on him (Gen 17:20; 21:13, 18). Paul’s κατὰ σάρκα is not a denial of paternity but a statement about the order under which the birth occurred. The rendering should not sharpen it into a slur (§5).

Built onmuraoka-lxx · bdag

NA28 prints the verse without a bracketed reading and the apparatus is thin. Two minor items:

• Some witnesses insert γεγέννηται (or a form of it) in the second clause to repair the ellipsis. The shorter, verbless text is original; supplying the verb is the obvious scribal improvement and the harder reading stands.
• A number of witnesses read τῆς ἐπαγγελίας (articular) for anarthrous ἐπαγγελίας, assimilating to Paul’s articular uses in ch. 3. The anarthrous qualitative form is to be preferred, and NA28 prints it.

Neither variant affects the sense; both illustrate the same scribal instinct toward regularization that the corpus policy on asyndeton and ellipsis is designed to resist.

Built onna28 · metzger-tc

The Latin renders the verse sed qui de ancilla secundum carnem natus est, qui autem de libera per repromissionem. Two observations: ancilla preserves the slave status, and repromissio (rather than plain promissio) reflects the sense of a formally reiterated undertaking — a reasonable reading of the LXX/Pauline force, though not evidence for the Greek beyond confirming the instrumental per.

Chrysostom, commenting on the passage, presses the point that the difference between the sons lies not in their father but in the mothers’ condition and the manner of the birth — and he draws the analogy to baptism as a birth not κατὰ σάρκα. Whatever one makes of the extension, his exegetical instinct is sound on the grammar: the single verb, doubly qualified, is what carries the contrast.

The patristic tradition also generally resisted using this verse to disparage Ishmael personally, treating it as typology about two covenantal orders rather than a verdict on a person — which is the same restraint §5 requires of the rendering.